CKA DEPLOYMENT & SERVICE LAB #2

Source: Dev.to

Global Rules

• Do NOT use kubectl edit
• Do NOT recreate resources unless asked
• Fix issues using kubectl patch / set / scale / rollout
• A namespace must be used
• Treat this like a real CKA exam

Baseline (Run on Every Machine)

Step 0 – Start cluster

minikube start --driver=docker

kubectl create namespace cka-lab
kubectl config set-context --current --namespace=cka-lab

Deploy the application

kubectl apply -f - <<EOF
apiVersion: apps/v1
kind: Deployment
metadata:
  name: web-app
spec:
  replicas: 4
  selector:
    matchLabels:
      app: web
  template:
    metadata:
      labels:
        app: web
    spec:
      containers:
      - name: nginx
        image: nginx:1.25
        ports:
        - containerPort: 80
EOF

Expose the application

kubectl apply -f - <<EOF
apiVersion: v1
kind: Service
metadata:
  name: web-svc
spec:
  selector:
    app: web
  ports:
  - port: 80
    targetPort: 80
  type: ClusterIP
EOF

Generate traffic

kubectl run traffic --image=busybox -it --rm -- sh

Inside the pod:

while true; do
  wget -qO- web-svc
  echo "-----"
  sleep 1
done

Tasks

Fix the deployment strategy

kubectl patch deployment web-app -p '
{
  "spec": {
    "strategy": {
      "type": "RollingUpdate",
      "rollingUpdate": {
        "maxUnavailable": 4,
        "maxSurge": 0
      }
    }
  }
}'

• No downtime during image updates
• Traffic loop never stops
• At least one pod always available

Simulate a failing image

kubectl set image deployment/web-app nginx=nginx:doesnotexist

Restore the deployment to a healthy state (without deleting)

kubectl patch deployment web-app --type=json -p='[
  {
    "op": "remove",
    "path": "/spec/template/spec/containers/0/readinessProbe"
  }
]'

• All pods Running, no CrashLoopBackOff
• Traffic loop remains stable

Ensure traffic only goes to ready pods (no cluster restart)

Patch the service selector to a wrong value:

kubectl patch svc web-svc -p '
{
  "spec": {
    "selector": {
      "app": "wrong"
    }
  }
}'

Restore correct selector (without recreating the service)

kubectl patch svc web-svc -p '
{
  "spec": {
    "selector": {
      "app": "web"
    }
  }
}'

• Endpoints repopulated, traffic resumes

Deploy a canary version

kubectl create deployment web-app-canary --image=nginx:1.27 --replicas=1
kubectl label deployment web-app-canary track=canary

• Canary receives traffic alongside the stable workload
• No service recreation

Remove the faulty canary immediately

kubectl delete deployment web-app-canary

• Stable version continues serving traffic uninterrupted

Pause rollout, update image, and finish rollout

kubectl rollout pause deployment web-app
kubectl set image deployment/web-app nginx=nginx:1.26

• Identify why the rollout is stuck
• Complete the deployment:

kubectl rollout resume deployment web-app

• New image fully deployed, rollout no longer paused

Expose the service via NodePort (no recreation)

kubectl patch svc web-svc -p '
{
  "spec": {
    "type": "NodePort"
  }
}'

• Application reachable from a browser via the assigned NodePort

Scale down to zero and restore service

kubectl scale deployment web-app --replicas=0

Restore the original replica count (do not recreate pods):

kubectl scale deployment web-app --replicas=4

• Pods run, traffic restored

Cleanup

kubectl delete namespace cka-lab

• Namespace must be empty

kubectl get all
# → No resources found